4ORIGINAL ARTICLE Cognition without a Neural Code: How a Folded Cortex Might Think by Harmonizing Its Own Electromagnetic Fields Victor M. Erlich, M.D., Ph.D. Department of Neurology University of Washington Northwest Hospital and Medical Center Seattle, WA ABSTRACT Extensive investigation of the brain’s synaptic connectivity, the presumed material basis of cognition, has failed to explain how the brain thinks. Further, the neural code that purportedly allows the brain to coordinate synaptic modulation over wide areas of cortex has yet to be found and may not exist. An alternative approach, focusing on the possibility that the brain’s internally generated electromagnetic fields might be biologically effective, leads to a model that solves this “binding problem.” The model of cognition proposed here permits mind and consciousness to arise naturally from the brain as trains of signifying states, or stationarities. Neuronal circuits in suitably constructed hierarchies produce thought by reconciling themselves with each other through the forward- and back-broadcast of specific electromagnetic fields, executing a natural algorithm as a harmonized set is selected. Beyond the postulation that information is encoded in specifically organized electromagnetic fields, the only other “code” necessary is topo- graphic, one that is already known. That the brain might use its own fields to think is supported by the literature on the widespread sensitivity of biological organisms to small, windowed fields. This model may help explain the coher- ence of the brain’s fields, the conservation of the folded cortex, and, in its emphasis on a self-harmonizing process, the universality of the esthetic impulse as a projection of the brain’s basic mechanism of thought. COGNITION BASED BOTH ON FIELDS AND ON SYNAPTIC of synaptic interconnectivity, and they argue that after CONNECTIVITY abundant “parallel processing” the brain somehow integrates what it has previously divided. But it is not ny credible model of thought must embody in clear how interlocking webs of neurons can represent a natural way the known physical features of one discrete “subunit” in an instant while suppressing the human brain. Such a model will encode a all other representations. An even deeper mystery is Amyriad of memory traces, and it will assemble how modification of synapses while thought is in prog- them rapidly and reliably, excluding extraneous matter ress can turn a mass of electrical activity into thought. without having to consult any list of rules, for which Synaptic change does not proceed at the rate humans there is no time or place. The model will automatically think, but over many seconds and even over days and focus attention where attention is needed, and it will months (Donoghue 1995). While synaptic modulation is spontaneously apply an internal logic that leads to sur- indispensible for learning, how does this learning actu- vivable behavior. At the same time, it will demonstrate ally help make thought? the human capacity for loose associations, self-contra- dictions, and even wild delusions. The model must be Even if brisk synaptic modulation were possible, say self-reading (no homunculi allowed), self-validating, by rapid functional changes rather than by structural and self-conscious. A model of human cognition should remodeling (which requires time-consuming protein perform as the brain does; it should be at once reliable synthesis), we are still left with the “binding problem,” and suspect, esthetic and erratic. the chief failure of cognitive neuroscience (Treisman 1996). The idea of synaptic modulation simply does not Most such models rest on synaptic connectivity. possess the power to explain how the brain synthesizes Disappointingly, the study of neurons and their connec- its “massively parallel processing.” Consider, for exam- tions in the cortex has produced much detailed physiol- ple, the work of Gerald Edelman (1987, 1992). He has ogy and many wiring diagrams but no convincing model built an admirable model of how different domains of of how the cortex makes an image, let alone a thought representation within the brain “map” themselves onto (Felleman and Van Essen 1991; Van Essen 1997). each other. The various domains stimulate each other and thereby mutually adjust, modifying their synapses Neuroscientists have demonstrated that the brain parses accordingly. Edelman calls his model the “theory of neu- its tasks into multiple subunits, all of which involve to- ronal group selection,” or TNGS, drawing on Darwinian and-fro axonal conduction followed by modification ideas of mutual accommodation. Through reciprocal 34 EJBM, Copyright © 2011 4ORIGINAL ARTICLE How a Folded Cortex Might Think by Harmonizing Its Own Electromagnetic Fields signaling along axons, synapses are fine-tuned, and a decide to swing or not, to craft and launch a swing, an group of firing units is selected. This selection provides impressive feat, given that the “what” (curve?) and the the material basis of thought. “where” (over the plate?) of vision are processed sepa- rately (Ungerleider and Haxby 1994). Because TNGS requires a plethora of axonal signaling from one unit to another it requires time. We can get a There is simply no known mechanism by which axonal rough idea of the time pressure on TNGS by considering messaging and synaptic modulation can go that fast, a batter trying to hit a baseball. How much time does even if we allow for functional rather than structural he have for central processing? A fastball pitched at changes. The nervous system contains both electrical 100 mph arrives at home plate in about 410 ms. Time is and chemical synapses: the former devoted to rapid needed for visual input: at a minimum the 33 ms required reflexes, the latter to slower, more modulated responses to glimpse and distinguish one numeral from another (Kandel et al. 1991). While electrical transmission across (Saarinen and Julesz 1991), if not the approximately 100 a synapse is almost instantaneous, the gap junctions on ms needed to register a visually evoked potential. On which this kind of transmission relies are not readily the output side, an expert needs about 160 ms to swing modified. In contrast, the chemical synapse, which can a bat. This leaves at most 200 ms for central processing be modified and thus serve as a basis for memory and (Endo et al. 1999; Regan 1997). With practice, a batter learning, requires a minimum of .3 ms of synaptic time, can train his motor cortex, refining synaptic structures often up to 5 ms. TNGS, which rests on sets of discrete over years, and he can guess pitches in real time, but he modulations of synaptic connectivity, inevitably requires cannot know a pitch’s type or location. He has 200 ms to much synaptic time, say 20 ms for five instances of mod- FIGURE 1: A three-dimensional neuronal circuit anatomically fixed by rosettes. According to the right-hand rule, this circuit will automatically broadcast a signature electromagnetic field, determined by the course of its current. The Einstein Journal of Biology and Medicine 35 4ORIGINAL ARTICLE How a Folded Cortex Might Think by Harmonizing Its Own Electromagnetic Fields ulation after round-trips between two sets of coordi- The model herein rests on the very fields recorded by nated neurons (a small number of round-trips, since any electro- and magnetoencephalograpy, with the novel one center is already modulated by third parties before hypothesis that complex brains not only broadcast these it receives news back from a center it just signaled). fields but read them as part of the process that produces In addition, there is the time spent conducting along thought. Because the brain’s patterned electromagnetic axons. Even if we assume optimal compaction (Cherniak fields depend on the connected neurons that generate 1994), the magnetoencephalographic evidence cited in them, the synapse also plays a role in the model. No the- support of TNGS shows coordination between areas sep- ory of the brain can be complete if it does not account arated by 10 cm or more (Gaetz et al. 1998; Srinivasan et for the brain’s connectivity. But, according to the view al. 1999). With a maximum conduction velocity in lightly taken here, no theory of higher brain function can be myelinated neurons of 8 m/s, five round-trips along 10 complete if it does not explain why the brain broadcasts cm of axon take 125 ms. Next, there is the membrane coherent fields that correlate with cognition. Lashley, constant, the time a neuron spends summing its multiple never persuaded that patterns of synaptic connectiv- synaptic inputs, another 8 to 20 ms (Kim and Connors ity alone could explain thought, failed to confirm any 1993; Shadlen and Newsome 1994) for each summation, theory of cognitive function based on field theories, say 100 ms for just five round-trips between two inter- including those of Kohler (Lashley et al. 1951). Lashley’s connected neurons with other inputs. The time require- failure convinced many that field theory is a dead end. ments of axonal conduction, synaptic transmission, and But experiments continue to demonstrate a relationship membrane constants exceed what is available to hit a between the brain’s thoughts and its fields (Freeman baseball, without any time spent on the synaptic modu- 2004a, 2004b, 2005; Freeman et al. 2003). lation that would be necessary to create the cortical rep- resentation of a discrete swing. TNGS works too slowly to explain how the cortex assembles its output, and so BIOLOGICALLY EFFECTIVE ELECTROMAGNETIC FIELDS do other models based on synaptic modulation and axo- nal conduction. In contrast, a model partially based on Azanza and del Moral (1994), who review the large sub- fields allows much of the brain’s computation to pro- ject of the biological effects of magnetic fields, conclude, ceed instantly. “The paramount result, in our opinion, is the ubiqui- tous response of biological systems, of any complexity, Further, the schema of integration in TNGS, abstracted to magnetic fields, which shows that magnetosensitiv- from its physical basis, is not sufficient to serve as a model ity is, indeed, a general phenomenon in living systems.” of thought.